What people are reacting to is the posed threat, or creepiness, of something. And what we should ascertain is whether being filmed is creepy. Being followed by a man not behaving according to social rules is an obvious possible threat that hardly needs to be proven, but how creepy it is to be surveillanced is another and more interesting point to prove.

So the experiment should be redone to factor out the creepy man. That means, either have a trusted and passive person do the filming (police, reflex-vest, etc) or to set up a tripod with a camera. Even if it was made clear that the video would be streamed to public, I think very few people would react as they did above, simply because a film camera is a lot less threatening than an unknown, weird person.

Dressing up the cameraman in different clothes and behavioural patterns would emphasize this point further. If he seem to be a tourist, a store employee, etc he would be judged less of a threat.

It's not double standard to judge creepy cameraman differently than CCTV, it's simply a rational conclusion that the cases are different.